# Meeting transcript cleanup prompt: free ChatGPT template > A copy-ready prompt that cleans a raw meeting transcript before you summarize it: fixes speaker labels, removes filler, flags unclear lines. Source: https://learn.ynteractive.com/content/prompts/meeting-transcript-cleanup-prompt · Updated 2026-10-12 · Free from Stepthrough (https://learn.ynteractive.com) [Meetings and notes](https://learn.ynteractive.com/content/prompts#job-meetings-and-notes) · [Accounting](https://learn.ynteractive.com/content/prompts#industry-accounting) · [Operations](https://learn.ynteractive.com/content/prompts#industry-operations) · Works in ChatGPT, Claude, Gemini and Copilot Updated October 12, 2026 · By [Robert Breen](https://learn.ynteractive.com/content/about) Use this on a raw transcript from Zoom, Google Meet or Teams before you ask for a summary. Automatic transcripts have wrong or missing [speaker labels](https://learn.ynteractive.com/content/glossary/speaker-labels), filler words and garbled names. Cleaning first gives the summary a better start. The prompt is strict about meaning: it may fix labels and remove "um," but it may not rewrite what anyone said. Anything unclear gets flagged for you, not fixed by a guess. ## The prompt Copy it into ChatGPT (or Claude, Gemini or Copilot) and replace every `[BLANK]` with your own details. It uses the four parts from [Prompt Writing 101](https://learn.ynteractive.com/content/prompt-writing-101-role-context-task-format): Role, Context, Task and Format. Copy prompt ```text Role: You are a careful note-taker at [BUSINESS], cleaning up a meeting transcript before it is summarized. Context: The meeting was [MEETING]. The people in it were [ATTENDEES]. Below is the raw transcript from the meeting tool. Names, numbers and dates may be misheard. Do not change what anyone said, and do not add anything that was not said. Task: Clean the transcript: put the right attendee name on each line, remove filler words and false starts, and fix obvious misspellings of the attendees' names and of [TERMS]. Where a line is unclear or a number or date may be misheard, keep it and add the note "(CHECK)" after it. Format: Keep the original order and the timestamps. One line per speaker turn, written as "Name: words". Plain text only. After the transcript, list every line marked "(CHECK)" in a short list with its timestamp. Here is the raw transcript: [TRANSCRIPT] ``` ## Fill in the blanks **`[BUSINESS]`** Your business name and what it does, in a few words. **`[MEETING]`** What kind of meeting and when, such as a Q3 review with a client on Monday morning. **`[ATTENDEES]`** Every person with their first and last name and role, so the labels can be fixed. **`[TERMS]`** Names and terms the tool often gets wrong, such as client names, software or jargon. **`[TRANSCRIPT]`** The raw text copied from the meeting tool. Leave out account numbers and tax IDs first. ## Example, filled in **A made-up example** from the free lesson [Turn a Client Meeting into Action Items and a Follow-Up Email (Accounting Firms)](https://learn.ynteractive.com/content/chatgpt-client-meeting-recap-accounting-firm). Millbrook Accounting is the made-up firm in the client meeting lesson. Maya wants the Q3 review transcript with Cedar Lane Landscaping cleaned up before she asks for a summary. Copy example ```text Role: You are a careful note-taker at Millbrook Accounting, a small accounting and bookkeeping firm, cleaning up a meeting transcript before it is summarized. Context: The meeting was a Q3 review with a client, Cedar Lane Landscaping, on Monday morning. The people in it were Tom Alvarez (bookkeeper), Maya Brooks (client manager), Grace Whitfield and Jordan Pike (client). Below is the raw transcript from the meeting tool. Names, numbers and dates may be misheard. Do not change what anyone said, and do not add anything that was not said. Task: Clean the transcript: put the right attendee name on each line, remove filler words and false starts, and fix obvious misspellings of the attendees' names and of W-9, uncategorized and reconciliation. Where a line is unclear or a number or date may be misheard, keep it and add the note "(CHECK)" after it. Format: Keep the original order and the timestamps. One line per speaker turn, written as "Name: words". Plain text only. After the transcript, list every line marked "(CHECK)" in a short list with its timestamp. Here is the raw transcript: (paste the transcript text here, without account numbers or tax IDs) ``` ## What a good answer looks like - Each line starts with a real attendee name instead of "Speaker 2", and the order and timestamps match the original. - Filler like "um" is gone, but a hedge such as "I think we can try Friday" stays, because it changes the meaning. - The "(CHECK)" list is short and specific, pointing at numbers, dates and names. ## How to check it before you use it An AI draft can sound right and still say something you never told it. Use the habit from the [Reply Faster lesson](https://learn.ynteractive.com/content/chatgpt-reply-to-customer-emails): check every promise, name, date and number against what you gave it before anything goes out. - Spot-check five lines against the recording. If any wording changed, tell the model to restore the original and rerun. - Resolve every "(CHECK)" yourself before the transcript goes into a summary. - Keep the raw transcript as the record. Use the cleaned one only as input for notes. - Follow your firm's policy on AI tools for client meetings, and remove account numbers and tax IDs first. ## Why this prompt works Separating cleanup from summarizing is a small [prompt chain](https://learn.ynteractive.com/content/glossary/prompt-chaining). The cleanup step is allowed to fix labels but not meaning, so errors are caught before they become decisions in a summary. ## Practice it in the free lesson **[Turn a Client Meeting into Action Items and a Follow-Up Email (Accounting Firms)](https://learn.ynteractive.com/content/chatgpt-client-meeting-recap-accounting-firm)**: Turn the transcript of a client's quarterly review into the decisions, an action items table with owners and due dates, and a follow-up email to the client. You do every step yourself in a practice copy of ChatGPT, and nothing touches your real accounts. [Start the free lesson](https://learn.ynteractive.com/modules/accounting-meeting-summary) Also useful: [Summarize a Meeting Transcript with ChatGPT](https://learn.ynteractive.com/content/chatgpt-summarize-meeting-transcript) ## Terms used here [Meeting transcript](https://learn.ynteractive.com/content/glossary/meeting-transcript) · [Speaker labels](https://learn.ynteractive.com/content/glossary/speaker-labels) · [Prompt chaining](https://learn.ynteractive.com/content/glossary/prompt-chaining) ## Related prompts - [Meeting recap email prompt](https://learn.ynteractive.com/content/prompts/meeting-recap-email-prompt): A copy-ready prompt that turns your checked meeting notes into a short recap email: decisions, owners and due dates, and nothing… - [Action items table prompt](https://learn.ynteractive.com/content/prompts/action-items-table-prompt): Paste a meeting transcript or rough notes and get a table of every task, owner and due date, with "Not stated" wherever nobody… ## More meetings and notes prompts - [One-on-one meeting prep prompt](https://learn.ynteractive.com/content/prompts/one-on-one-meeting-prep-prompt): Prepare a one-on-one from your notes: what you owe them from last time, specific wins, questions that let them talk first, and… - [Client kickoff agenda prompt](https://learn.ynteractive.com/content/prompts/client-kickoff-agenda-prompt): Build a kickoff meeting agenda for a new client project: introductions, goals, scope, who does what, timeline and the first… - [Client quarterly review recap prompt](https://learn.ynteractive.com/content/prompts/client-quarterly-review-recap-prompt): Turn your notes or transcript from a client's quarterly review into a recap email: decisions, who does what by when, what you… ## Frequently asked questions **Should I clean a meeting transcript before summarizing it?** Often yes, if the labels or names are wrong. Wrong labels lead to wrong owners in an action items table. For a short, clean transcript you can skip this step. **Will AI change what people said when it cleans a transcript?** It can, which is why the prompt forbids it and asks for "(CHECK)" flags. Spot-check a few lines against the recording. [All free prompt templates](https://learn.ynteractive.com/content/prompts) · [AI glossary](https://learn.ynteractive.com/content/glossary)